Article — "Precision Senolytics: Circulating ANGPTL8 and SASP Spread Reveal New Druggable Targets": Article — "Precision Senolytics: Circulating ANGPTL8 and SASP Spread Reveal New Druggable Targets": A mouse with the gene Angptl8 deleted, at 22 months of age—old, by murine standards—hangs onto a spinning rotarod longer than its wild-type littermate, runs farther on a treadmill, explores more, carries more gastrocnemius muscle with a higher fraction of fast-twitch fibers, and ultimately outlives it.
No one gave that mouse a senolytic. No senescent cell was killed.
The animal aged better, reportedly because one circulating protein—produced mainly by senescent fat cells, and reportedly rising with age in mice just as it does in people—was simply never there (E1). Its fat tissue reportedly carried less senescence. Its blood reportedly carried fewer SASP factors (E1). That single deletion propagated outward into function and lifespan (E1).
This is the quiet turn the senolytics field has been waiting for, and it inverts the entire premise. The dominant strategy has always been clearance: find the senescent cell, kill it, let the damage stop. The Angptl8 result says you may not need to touch the cell at all. You can drug the signal it sends.
The dominant strategy has always been clearance: find the senescent cell, kill it. The Angptl8 result says you may not need to touch the cell at all.
Call it effector-targeting. Instead of hunting a heterogeneous, hard-to-mark cell population, you neutralize a specific molecule that carries senescence's damage into the rest of the body.
The case for why this matters begins with why clearance is dangerous in the people who most need it.
Senescence is not a villain with a single face. It suppresses tumors and heals wounds when transient; it drives inflammation, fibrosis, and age-related disease when it lingers. Scott Lowe and colleagues frame the translation problem bluntly: senescent-state heterogeneity, limited biomarkers, off-target toxicities, and—critically—the frailty of the aged patients you'd be treating (E4).
The workhorse senolytic exposes the trap. Navitoclax kills senescent cells by inhibiting BCL-2 family proteins, but it also depletes the BCL-XL that platelets need, and at effective doses it reportedly causes thrombocytopenia—a condition that can lead to uncontrolled bleeding. It remains unapproved by the FDA (E3).
Worse, broad BCL-2 depletion doesn't kill all cancers; against some lines it can stabilize the cancer cell instead (E3). You are asking a frail patient to accept bleeding risk for a drug that, in the wrong tumor context, might help the disease it's supposed to hedge against.
Now consider how heterogeneous the target really is—and here the honest complexity of clearance comes into focus.
When researchers profiled the SASP across five human brain cell types—astrocytes, endothelial cells, microglia, oligodendrocytes, neurons—there was no shared "senescence signal." A panel of 268 cytokines showed each cell type broadcasting its own combination, with only a small handful common to all (per E6; E2 states 286).
The direction of spread was just as idiosyncratic. Microglia could be driven senescent by almost any conditioned media except that of endothelial cells; neurons and oligodendrocytes resisted secondary senescence entirely, no matter the source (E2).
The single most damning detail for the "one off-switch" fantasy: astrocyte SASP applied to astrocytes raised CDKN1A, the gene encoding p21, significantly. The same astrocyte SASP applied to microglia lowered it (E2, E6).
Same signal. Opposite response. There is reportedly no universal senescence lever to pull—which, per E2 and E6, would mean a drug that shuts down "the SASP" is a category error before it's even a molecule.
What the brain study offers instead is a menu of specific ligand-receptor pairs: CCL2, MIF, CXCR7, DPP4, named as candidate targets. The CCL2 inhibitor Bindarit produced significant—if incomplete—reductions in senescence transmission between astrocytes (E2, E6).
That's the effector-targeting logic in miniature: not "clear the cell" but "block this factor, in this direction, between these cell types." The catch is scale—this was cell culture only, five cell types driven senescent by a chemical, conditioned media that don't reflect the real brain's factor mix. No animal, no human (E2, E6).
Even the field's more sophisticated clearance attempts can't fully escape the killing paradigm. The Conboy lab's DMA cocktail—dichloroacetate, metformin, and navitoclax—is clever precisely because it exploits a shared vulnerability: senescent cells and cancer cells both run on broken mitochondria, produce less ATP, and lean on glycolysis (E3).
DMA starves them of energy. At what E3 describes as roughly one-tenth the usual navitoclax dose—1 μM, which E3 compares to the 50 mg/kg standard used in mouse studies—it killed roughly 60% of damage-induced senescent cells while sparing healthy ones, and in a separate experiment, unlike full-dose navitoclax, it reportedly didn't significantly drop platelet counts (E3).
That's a real safety gain. But notice the strategy still leans on low-dose navitoclax to broadly kill; it is a gentler clearance, not a departure from clearance. And in aged mice, DMA improved treadmill performance while leaving hanging tests and overall frailty unaffected (E3).
Against that backdrop, ANGPTL8 looks like the cleanest target the field has produced—because the case is built from both directions at once.
Mechanistically, ANGPTL8 binds AKT2 and drives the AKT/mTOR/S6K axis; knockout mice show decreased activity across all three, and experiments confirmed AKT is required for ANGPTL8 to push cells toward senescence (E1). The protein is not just a marker. It's a functional effector.
The human correlational data are strikingly consistent. In a biological-age clock built from nearly ten thousand people in a Chinese cardiometabolic cohort, circulating ANGPTL8 near 250 ng/L mapped to the lowest biological age—while people carrying four times that level or more scored markedly older (E1).
Remove ANGPTL8 from the clock and its predictive power collapsed. The researchers call it "a consistent, biologically meaningful contributor to age prediction across individuals" (E1).
The mortality signal sharpens it. In the REACTION study of diabetics, those in the top ANGPTL8 quartile reportedly carried a 3.54 risk ratio for death versus the bottom quartile—and the relationship was linear (E9). A separate REACTION analysis, based on novel diabetes subgroups rather than an identical cohort, reportedly found a comparable hazard, a risk ratio of 3.23 for all-cause mortality across a quartile 4 vs 1 split (E10).
The researchers' own framing captures the appeal: ANGPTL8's "dual identity" as both circulating biomarker and functional effector of aging is what makes it "particularly compelling" from a translational standpoint (E1). It is a signal you could, in principle, mop up without ever killing a cell.
Here is the honest last step, and it is unbuilt.
The knockout mice never had ANGPTL8 in the first place—genetic absence from conception, not therapeutic removal from an aging animal. The paper did not test any method of clearing circulating ANGPTL8 from the bloodstream of a normal, wild-type animal (E1).
That is the entire promise of precision senolytics condensed into a single missing experiment. We have a mechanistically validated effector, a dose-response human clock, and a hard mortality association—and no drug, no antibody, no evidence that removing this protein from a living aged body reproduces the knockout's gift.
And every human number is correlational. ANGPTL8 travels with hypertension and other established risk factors; the researchers themselves caution it "may amplify age-related clinical vulnerability rather than acting solely as an isolated risk marker" (E1). Amplifier or bystander is not yet settled.
So the strategic argument stands even as the deliverable does not. Effector-targeting sidesteps the two traps that make wholesale clearance dangerous in frail patients—the platelet toxicity of BCL-2 inhibition and the context-dependent cancer risk of removing senescence's tumor-suppressive function—because it never touches the cell's survival machinery or its protective role at all.
The brain-SASP web proves there is no single cell to clear cleanly. ANGPTL8 proves there may be single signals worth silencing.
Precision senolytics, right now, is a promissory note. The most translatable idea in the field is not a better way to kill—it's the recognition that you might not have to. What's missing is the one drug that would turn a longer-lived knockout mouse into a treatment.